Using multiprocessors

I'm running Skeinforge on Mac Leopard and I notice that it only uses one CPU at a time. Is there a way to use more to speed it up? It's a shame to waste all that processing power.

from enrique on his blog:

Quote

> any chance of ever introducing some
> parallelism into skeinforge?..

Not for most of skeinforge. Skeinforge does a lot of stuff sequentially and it often looks at the layers around the layer it is generating. The only parts that could be parallelized with reasonable difficulty would be the in the carve tool. However, I do not currently plan on doing this.

To speed up skeinforge, people have had some success with pypy:
[codespeak.net]

and psyco:
[psyco.sourceforge.net]

I also read that someone ran skeinforge on several objects in parallel. To do this, I believe they made a shell script and ran an instance of skeinforge for each object. Unfortunately, I couldn't find the post mentioning that.

Thanks. I tried to doing a quick search but I came up empty.

Yes. Doing several objects in parallel would be easy but then you still have to combine the g-codes layer by layer. Not too hard but doable with a script. I would be interested in seeing that script.

Within skeinforge, it can certainly run separate processes when objects divide into multiple isolated "islands" or towers. It can also divide an object along z and do a full fill at the beginning of each segment. Then it can send each segment to separate processors.

Anyway, Skeinforge is probably the most stable and most effective of all the SW that I'm using with the Mendel right now. But, it would be nice if it can go faster with large parts or when printing a full tray of parts. One of the parts that I'm setting up to print has 4 pieces that are up to 130 mm tall and it takes an hour or so for Skeinforge to convert to g-code.

wow that's absurd. are you using the speed-ups mentioned?

No. But,if I had a way to combine the g-code files, I would run each piece separately through Skeinforge as you wrote. It's a piece that's a little over 400 mm long so I split it into four parts which plan to glue together.

I finally got the chance to install psyco on Mac Leopard and I've been seeing a 2x speed improvement. This definitely helps. Thanks!